I`m lost. Totally awfull measurements in my room. Need advice.

[kyle_neuron]

Hmm, that’s a little more off than I was expecting. The green trace also doesn’t align mathematically with the traces you uploaded for each box solo.

How is your sub connected to the rest of the system? Can you draw a diagram?

Also, when you did the solo traces, what position was the sub ‘phase’ switch set to?

 

[Bear123]

Hi there,

I bought a UMIK-1 and tried to measure my Speakers in REW and I`m getting a very uneven frequency response. I used the EQ function and it threw a dozen of filters that are all of very high Q value and massive volume change.

Can you maybe have a look if this all makes sense? Not sure, if I`m doing something wrong. The algorithm reduced the volume by more than 12db in some areas. Are my speakers really crap or do you think the room can "destroy" the sound that much?

Here`s the before and after measurement:

View attachment 152876

And this is the implemented EQ which the "after" mearuement is done with with a litte manual optimization above 600Hz

View attachment 152877

Really not sure if all of this makes sense. The sound changes drastically and I think definitely to the better. But it looks sooo radical.

Is there anything I could do to my room to lessen this rollercoaster bewtween 100 and 400Hz?

Sorry if these are dumb questions, but I really don`t know what to think about it.

Just a good example of why "pure 2.0" is nowhere remotely close to high fidelity.

Looks like you have a couple of issues well below 100 Hz which can be addressed either by sub placement, or adding another. I'd try moving the sub around first to see if you can eliminate or lessen the large dips in response that eq cannot fix.

 

[Wegi76]

Hmm, that’s a little more off than I was expecting. The green trace also doesn’t align mathematically with the traces you uploaded for each box solo.

How is your sub connected to the rest of the system? Can you draw a diagram?

Also, when you did the solo traces, what position was the sub ‘phase’ switch set to?

The sub is connected wirelessly trough a proprietary link system (Nubert NuPro X-4000 RC & NuSub SW900) I can control the sub via a smartphone app and set volume relative to the mains, phase 0° or 180°, low-pass frequency in 1 Hz steps. The mains do have the correspnding high pass filterfrequency setting. So, the sub is only conneted to the main line. I unfortunately cannot run the sub seperatly over the wireless link and have to use it`s Aux-In in this case. The mains include a DAC and are connected via USB to the computer.

On all solo traces the subwoofer phase is set to 0°. I did not create traces with the subwoofers phase inverted so far. Should I?

 

[Wegi76]

@Bear123 I tried that. The current placement in the front right corner is already the result of these tests. The worst dip @33Hz is not so critial to my ears.

 

[kyle_neuron]

The sub is connected wirelessly trough a proprietary link system (Nubert NuPro X-4000 RC & NuSub SW900) I can control the sub via a smartphone app and set volume relative to the mains, phase 0° or 180°, low-pass frequency in 1 Hz steps. The mains do have the correspnding high pass filterfrequency setting. So, the sub is only conneted to the main line. I unfortunately cannot run the sub seperatly over the wireless link and have to use it`s Aux-In in this case. The mains include a DAC and are connected via USB to the computer.

On all solo traces the subwoofer phase is set to 0°. I did not create traces with the subwoofers phase inverted so far. Should I?

Does the wireless link run via the main speakers? Or is it via a dedicated processing box?

What I’m trying to figure out is if you feed a left and right signal from your PC to *something* and that then passes it onto the sub. If so, and your computer is only outputting stereo, your delay on the right channel will be reaching the sub as well.
Edit: you clearly answered this and I need to read properly

In response to the subwoofer polarity switch, my predicted settings were done with the mains both at inverse polarity to what the sub was set to in the solo measurements. So to get the results I predict, you’d need to have the sub polarity switch at the 180-degree position. That puts it in phase through crossover with the left main speaker.

The wireless system is also a concern, as it will have it's own latency which we need to compensate for. If you fed the sub via a cabled connection for measurements, then we don't know or include that latency when looking at creating an alignment. Unless the mains run via the same wireless platform, or the software guys were smart enough to latency compensate it in their digital processing, that is.

You'd really need to feed the sub separately from the music source PC, so that it doesn't receive the delayed signal of the right main speaker.

Forgive me if this is different to how the system works. I'm not familiar with that brand of speaker.

The alternative of course is to move the sub physically to change the timing offset. Right now, it looks to be closer to one side of mains than the other, but will also have a 'delayed response' due to the cycles it's reproducing. That's why we're trying a 'depth shift' using delay on one main speaker, to align on a virtual horizontal plane to the 'later" source.

Hope that makes sense.

 

[Wegi76]

You are absolutely right. REW generates sweep --> EQ & Delay for L/R in Windows (Equalizer APO)--> USB to speakers as 2 channel pcm data

-->a) D/A conversion and output on the mains with independent channel information
-->b) wireless link of mono downmix to the sub --> low-pass, D/A conversion and output

Since the speakers have an analogue input too, I could use this too and feed it from my RME DAC. But this will also send a summed mono signal, so with delay. (see page 14 of the manual)

https://www.nubert.de/downloads/nupro-x-4000-rc-manual-e12-en.pdf

Don`t think I can do this in Windows at least not with Equalizer APO, since its a global EQ. Think I`d have to split the signal into two stereo signal chains before EQ/Time correction and have two separate outputs. The two outputs are not the issue, but I don`t know how to separate the signals. (hope not to think to complicated...)

Does maybe anyone else have an idea? Would a miniDSP 2x4 do the job if there is no software solution?

The alternative of course is to move the sub physically to change the timing offset. Right now, it looks to be closer to one side of mains than the other, but will also have a 'delayed response' due to the cycles it's reproducing. That's why we're trying a 'depth shift' using delay on one main speaker, to align on a virtual horizontal plane to the 'later" source.

Yes, it sits in the right corner and therefore closer to the right main. Here`s a drawing

 

[GalZohar]

I don't understand half of what was explained here, but what I do know is that if you got better (but not optimal) response flipping the subwoofer phase, then what you really need is to adjust the delay of the subwoofer (which will then affect the phase in a continuous manner), rather than just flipping the phase.

Usually done after applying EQ separately to speakers and sub to get the best response for each separately, then set and apply the crossover, then measuring everything together you adjust subwoofer delay to find the one that will give you the flattest response for the combined subwoofer+speakers with all EQ and crossover applied. I use Audyssey so the EQ is automatic and supposedly so is the subwoofer delay, but in practice the subwoofer delay set by Audyssey was absolutely bad, and had to almost flip the polarity by adding 2m of subwoofer distance in my receiver (+2.18m would have been a full 180 degrees at 80Hz, which gave very slightly worse response compared to +2m). For me the area between 60Hz and 100Hz was greatly improved by just this change.

If you have a simple way to just add change the delay of the subwoofer, that alone can give you a significant benefit, especially considering the improvement you already got by just flipping the phase, which gets you closer to the phase you need, but not exactly where it needs to be.

Supposedly you still can't align frequencies far from the crossover frequency (except maybe with advanced systems I have no experience with), but they also don't matter nearly as much, as far from the crossover frequency most energy comes from either subwoofer or speakers and any phase difference between them is insignificant when one overpowers the other.

 

[kyle_neuron]

Phase alignment is about relative adjustments, so you’re on the right track but in most cases, the sub will be the ‘latest arrival’ and therefore is rarely the unit that needs further delay adding to it - assuming the mains and sub are on the same horizontal plane.

In this case, the traces show the sub is already the later arrival when set with the polarity switch at 0 degrees. A polarity reversal is not the same thing as adding delay to a system, but it can help to reduce the amount of delay needed to get a result.

The general process is to measure all units solo with the mic at a given location. The latest arrival is the one with the steepest phase trace at the desired transition region. The shallower traces need delay or filter changes to match the slope of the latest arrival. If you get the same phase trace shape, but they don’t overlap, then apply a polarity flip to get you there. All-pass filters are an advanced way to do this, if available.

But it takes a bit of practice to get good at this. Especially in rooms, where there’s so little data in the FFT at the low end, and reflections come into play. That’s where an unwrapped minimum phase plot can help.

The maths is then fairly simple.
T_delta = Ph_angle / (360 * f)

Where f is frequency.

A worked example; if you have a sub trace and the main speaker trace with a measured 40 degree offset at 100 Hz, and it's 'normal' conditions of 20 degrees C then you end up with a 1.111 ms delay needed to align the slopes at that frequency.

It might not be aligned above and below, depending on the filter slopes in use, position, etc. You're aiming for a good, not perfect though. Anything within a 30-degree corridor of overlap is a summation. Anything within 90-degree is less gain from summation, down to nothing extra.

@Wegi76 a MiniDSP would work, but I think they're a little noisy personally. It would be a shame to add further A/D and D/A conversions to your setup - do the mains have a digital input other than USB? Perhaps AES3 or SPDIF?

 

[Wegi76]

Hi guys, first of all many thanks for your great support, especially to Kyle!

I am also concerned be the MiniDSPs DAC quality. The hardware is really old...hmm...

I think there is currently no workaround to the fact, that I only have 2.0 output in my system for a pure software solution. Some guy wrote a huge manual how to to do multichannel correction within Windows and with additional tools like Voicemeter and so on.... But you need a multichannel soundinterface to make that work as a basic prerequisite.

Check: https://www.audiosciencereview.com/...-a-8-ch-pre-pro-experiment.14785/#post-462671

Anyway... The good thing about the "wireless" sub is, that even if I can`t set delay, I can continue to move it around in the room without too many cabeling issues. Will concentrate on that now and do some more try and error.

Also I downloaded Dirac Live as a trial and gave that a try. It also calculated a delay to the right speaker (0.1ms) and the overall frequency response is a little better than my manual compensation. But it`s not worth 250€ for me, since the REW -> EQ Plugin already delivered >90% of the results DIRAC calculated. (With 500% more time and effort involved to be fair...)

So, even if there is no happy end so far I learned a alot from you. 1000x thanks for that!

Cheers
Markus

 

[Wegi76]

do the mains have a digital input other than USB? Perhaps AES3 or SPDIF?

Sorry, forgot to answer: Yes they do: I have a AES/XLR input, SPDIF, digital coax, RCAs, USB (and Bluetooth)

 

[kyle_neuron]

Well that becomes more interesting. You could use VoiceMeeter to make a combined virtual interface if the computer has a SPDIF output as well as analogue.

Of course, as you say you can also try bringing the sub forward into the room, which would achieve similar results. Hopefully the tweak to your REW process will help with that!

You’re using the UMIK USB mic, right? The downside with those is that they prevent dual-channel measurement with loopback for calculation of timing offset. So long as you keep the mic in the exact same position, it’s not an issue, but having the loopback does speed things up when learning.

A Behringer UMC202HD and their cheap measurement mic is a useful thing to have. Or go for the 404HD and multiple mics… you can probably make back the €200 you’d spend by using your new skills to fix your buddy’s hifi systems

 

[Wegi76]

Hi

I actually played with Voicemeter and E-APO as an bus insert. (I have found a multichannel Soundcard somewhere in the house - Soundblaster AE7) In E-APO I can then EQ and Delay all of these 5.1 channels individually in theory. But I failed to send the fullrange signal to let`s say the rear speakers output to which the sub would be connected. I think this is because my signal is still 2.0 and the bloody thing expects a true multichannel encoded signal. Probably another dead end here...

And yes, its a UMIK-1. Sorry if that`s a dumb question: But why does`nt the acoustic timing reference work here? Guess I still have much to learn.

If money did not matter, the MiniDSP SHD looks great too:
+ Very good DAC, no hiss to be expected
+ DIRAC (full version) although only 2.0 and no bass management
+ All the MiniDSP crossover functionalities --> no problem to add the sub and control it individually regarding phase, delay, equalization
- ~1500€

 

[zermak]

@Wegi76
You can use either Voicemeter virtual in or a real 5.1/7.1 channels sound card to have access to the multiple outputs and put the fronts (L+R) outputs on any channel you like.
In Equalizer APO you have to copy the L and R channels into the channel that you want to have the outputs. Best way is to copy half of each signal from L and R and output it on the SUB out (or any output you want). I use it myself. Here is the simple code I use:

Copy: SUB=0.5*L+0.5*R

 

[zermak]

Another thing. I advise you to use the chanlles selection filters to control the EQs on each channel (L, R and SUB in this case). I have read you had trouble to set a single delay to a single channel; You just need to select the channel you want to add delay to and then use the delay thing.
The code should look something like:

Copy: SUB=0.5*L+0.5*R
Channel: L
# any EQs for L channel
Channel: R
Delay: 3 ms
# any EQs for R channel
Channel: SUB
# any EQs for the SUB channel

The example above adds 3 ms to right channel just to show you how it looks like in the code.

I even use 48dB/oct high pass and low passes for the mains and the subs. Here is the code for a 24dB/oct low pass/high pass filters set at 80Hz respectively (NB: you just have to use two of this filter to get the 48dB/oct filter):

# LP 24dB/oct 80Hz
Eval: N = 4
Eval: Fc = 80
Eval: Q=1/(2*sin((pi/N)*(0+1/2)))
Filter 5: ON LPQ Fc `Fc` Hz Q `Q`
Eval: Q=1/(2*sin((pi/N)*(1+1/2)))
Filter 5: ON LPQ Fc `Fc` Hz Q `Q`

# HP 24dB/oct 80Hz
Eval: N = 4
Eval: Fc = 80
Eval: Q=1/(2*sin((pi/N)*(0+1/2)))
Filter 5: ON HPQ Fc `Fc` Hz Q `Q`
Eval: Q=1/(2*sin((pi/N)*(1+1/2)))
Filter 5: ON HPQ Fc `Fc` Hz Q `Q`

You can save each code in a txt file and use the include configuration file option to make the main page less messy

 

[Wegi76]

Oh man... A classic case of RTFM regarding your first advice. The solution was so near and I simply overlooked it. So thank you very much, zermak!

Regarding your crossover code... Woah... Did not know that EPO can do such things at all :-o Impressive...

Really messed up my system with all these interfaces, Voicemeter, Dirac drivers, Soundblaster, Realtek, ASIO4all and VB-Cable stuff. Nothing seems to work properly right now. Within the tone generator in REW (Java driver) it loooks like I got the routing right with your Copy command example and the sub responds to the "copied-to"-channel as expected but system sound is broken. Might have to clean up the mess first before going on.

Anyways, this is great fun and I won`t give up until I mange to add this bloody 3.29ms delay to my right speaker only

On the long run the MiniDSP SHD is still on my roadmap. Could benefit from room correction from different sources that way (built-in BT, my Macbook, Node 2i Streamer...) But it`s soooo expensive.

 

[zermak]

Oh man... A classic case of RTFM regarding your first advice. The solution was so near and I simply overlooked it. So thank you very much, zermak!

Regarding your crossover code... Woah... Did not know that EPO can do such things at all :-o Impressive...

Really messed up my system with all these interfaces, Voicemeter, Dirac drivers, Soundblaster, Realtek, ASIO4all and VB-Cable stuff. Nothing seems to work properly right now. Within the tone generator in REW (Java driver) it loooks like I got the routing right with your Copy command example and the sub responds to the "copied-to"-channel as expected but system sound is broken. Might have to clean up the mess first before going on.

Anyways, this is great fun and I won`t give up until I mange to add this bloody 3.29ms delay to my right speaker only

On the long run the MiniDSP SHD is still on my roadmap. Could benefit from room correction from different sources that way (built-in BT, my Macbook, Node 2i Streamer...) But it`s soooo expensive.

Yes, there is not much on EqualizerAPO page, just some basics examples but the software itself is very powerful.
I know all of this because I was in your situation and searching (some topics in the github page/forum helped a little... I got the above codes for the high and low passes filter there indeed) and experimenting was the only way to make it work properly (I even use HeSuVi for my headphones and messing up with it's code was kinda helpful and educational).
Sadly EqualizerAPO only works on DS/WASAPI shared or it would be awesome to use it with WASAPI exclusive mode and mainly ASIO to be able to use it in practically any audio software.

You only need the virtual audio card if you don't have a sound card/DAC with more than stereo outputs to fool the system and be able to use the eight virtual channels. I use it myself with my stereo DAC for my headphones to have virtual surround: to mix the 5.1/7.1 audio tracks of the TV shows/movies I watch using HRTF (head related transfer funcion) filters.

I think for your needs EqualizerAPO is the best all in one solution. Of course if you need EQ outside of your PC then you need hardware that does it, like some MiniDSP products.

 

[Wegi76]

It`s frustrating... I reinstalled the messed up Windows machine and started over again. I`ve come to the point, where I cannot measure anything in REW in the desired 3-channel-setup.

A) The analogue output of hte Soundblaster AE7 distortes when using the Java driver and
B) I can only access L/R channels, not the third channel with that.

Selecting the soundcardsASIO drivers I can select the output channels 1-6 BUT the UMIK-1 does not support ASIO. So, quickly googled for a solution and installed ASIO4All to create some bridge. But as soon as I select this as my Asio driver REW crashes. This seems to be a common problem with Asio4all.

Equalizer APO is really cool and the CRAVE EQ ist a great product too and I`ll continue to use these tools for my headphones, but I kinda lost my motivation as I spent many hours without success.

Hate to give but, but I did and ordered a Minidsp SHD. Dirac Live is only implemented for two channels but the backend-processing will finally enable setting Delay and Phase per channel independently.

At least I found time to install the Basotect panels mentioned earlier (not yet covered with cloth)